Influence of polyvinylpyrrolidone (PVP) in the synthesis of luminescent NaYF4:Yb,Er upconversion nanoparticles

Polyvinylpyrrolidone (PVP) can be used to produce upconversion nanoparticles (UCNPs) in an advantageous manner, i.e. at modest temperatures in open-to-air conditions with simple hotplate and flask apparatus. However, the influence of PVP parameters on the formation of UCNPs has not been previously investigated. In this exploratory study, we establish that PVP molecular weight and relative amount of PVP can greatly influence the morphology and diameter of NaYF4:Yb,Er UCNPs produced via the PVP-assisted route. At nominal amounts of PVP, varying the molecular weight of PVP in synthesis between 10,000 g mol−1 (PVP10), 40,000 g mol−1 (PVP40), and 55,000 g mol−1 (PVP55), had minimal effect on UCNP morphology, whereas reducing the quantity of PVP10 and PVP40 in the reaction to 10% of the nominal amount resulted in two notable effects: (1) the generation of a greater range of UCNP diameters and (2) the production of an unexpected sub-population of rhombus-shaped UCNPs. Bulk and individual nanoparticle analysis indicates that all UCNP morphologies were cubic (α-phase) crystal structure and consisted of NaYF4:Yb,Er. Optical emission properties exhibited only modest green and red luminescence emission ratio when PVP parameters were varied. However, separately produced PVP40 NaYF4:Yb,Tm UCNPs exhibited a much more intense and dual-band blue/red emission. This exploratory work demonstrates that tailoring PVP content in synthesis of UCNPs can greatly alter morphology of UCNPs produced and should be carefully considered in experimental design. However, the underlying mechanisms of action of the role PVP plays in this synthesis remain unclear. Ultimately, significant further work is still required to fully elucidate the relevant chemistry to achieve full control of PVP-UCNP synthesis.


Introduction
Polyvinylpyrrolidone (PVP) (see figure 1(a)) is a versatile polymer which fulfils a variety of purposes in colloidal nanoparticle synthesis: it is a reducing agent, it can control nanoparticle growth, and it can stabilize nanoparticle surfaces. The PVP monomer contains both a hydrophobic alkyl group and a hydrophilic pyrrolidone moiety, enabling PVP-capped nanoparticles to be dispersed in a wide range of organic solvents and water [1]. PVP has been utilized in the synthesis of a wide range of nanoparticles and nanostructures, e.g. Koczkur et al (2015) note the use of PVP in the synthesis of Au, Ag, Pd, Pt, bimetallic, metal oxide, and metal chalcogenide nanostructures [1]. Varying PVP molecular weight and relative amount in synthesis has been demonstrated to alter the morphology of nanoparticles produced, e.g. by encouraging growth along specific crystal facets [1,2]. Koczkur et al (2015) also noted that manipulation of PVP in synthesis can give rise to a wide variety of nano-morphologies, including nano-wires, cubes, plates, arrow-headed tripods, nano-flowers, nano-cages, tetrahedra, hexagon, polyhedral, and triangular morphologies [1].
researchers new to the field. For example, solvothermal and hydrothermal reactions require a high-pressure sealed autoclave capable of withstanding elevated pressures at temperatures typically up to 200°C, whereas 'hot injection' methods require oxygen-free reactions at ∼300°C for extended reaction times. These reaction conditions are hazardous and warrant skilled operation and careful risk mitigation. Furthermore, these techniques are very sensitive to variables such as reaction time, pH, and the ratio of solvents/ coordinating ligands used [12]. Therefore, reproducibility between different research groups can be challenging [34]. Ultimately, despite their near ubiquity in the field, these solvothermal/hydrothermal synthesis and hot injection synthesis are specialist synthesis methods, which are resource, expertise, and energy intensive, which are a barrier to entry for researchers, and which are ultimately challenging to scale-up [35]. Microwave-assisted synthesis provides an attractive route for straight-forward UCNP synthesis, with excellent prospects for in situ monitoring and batch reactions, but this requires specialist microwave reactor equipment [36][37][38]. Notably, some methods for producing UCNP materials at low temperatures (<50°C) with simple equipment have been reported, but these studies only report upconversion properties in dry solid or dry powdered states [39,40]. Therefore there is a need for a more straight-forward and costeffective method to synthesise water and solvent-dispersible UCNPs, especially for large-scale applications such as security inks, solar cells, and display technologies [17]. The PVP-assisted UCNP synthesis route provides a solution to these issues.
The PVP-assisted route has sporadically attracted attention since its innovation by Li and Zhang in 2006 (see figure 1(b)) [3]. It requires only modest-temperatures (160°C), a single solvent, can be conducted open-to-air, and requires only simple hotplate and beaker apparatus. As such the PVP-assisted method can be deployed easily, and is a promising candidate for directly scaled-up or batch-parallelised UCNP production, which would be particularly advantageous for applications requiring bulk quantities of  Li and Zhang (2006) [3], where Ln 3+ ions (Y 3+ ,Yb 3+ ,Er 3+ ) associate the PVP prior to the addition of F − ions, which leads to formation of PVP-UCNPs.
In the PVP-assisted method, it is speculated that the various trivalent lanthanide ions nucleate to the pyrrolidone moieties of PVP (see figure 1(b)) [1,43], with the subsequent addition of F − ions causing the formation of the NaYF 4 :Yb,Er crystalline lattice. PVP then acts to constrain the size of the resultant UCNPs, and to aid dispersal in solvents. Typical UCNP reaction time is 2 h and produces ∼130 mg of PVP-UCNPs. To date, red/green (Yb 3+ ,Er 3+ ), strong red (Yb 3+ ,Er 3+ ,Mn 2+ ), and blue emitting (Yb 3+, Tm 3+ ) UCNPs have been produced with the PVP-assisted route [3,17], but other emission wavebands may be accessible by incorporating different dopant combinations [18].
The PVP-assisted UCNP production method has been utilized to produce cubic (α-phase) NaYF 4 :Yb,Er UCNPs for various purposes, including sub-cellular microscopy, drug delivery, MRI imaging, and light harvesting in solar cells [5,6,29,44]. Typically the NaYF 4 :Yb,Er UCNPs which are produced using a nominal quantity of PVP40 in the synthesis (i.e. 574 mg) are approximately 40 to 50 nm in diameter (depending on diameter quantification method), exhibit a ∼1:1 emission ratio for red/green Er 3+ emission wavebands (see figure 2 for a corresponding energy level diagram), and a crystal lattice parameter, a, of 5.52 Å which has been reproducible between studies [6,17]. Furthermore, PVP-UCNPs can be functionalised with silica shells to mitigate solvent quenching or mesoporous silica shells for functional molecular loading [3,5,45].
Recently, our group has established that the PVPassisted method can produce cubic (α-phase) NaYF 4 :Yb,Er,Mn UCNPs with remarkably enhanced red emission [17]. The sixfold increase in red emission was attributed to incorporation of Mn 2+ co-doping resulting in (1) a reduced crystal lattice parameter (i.e. from 5.52 Å to 5.45 Å); (2) three-photon sensitized upconversion; and (3) reduced nanoparticle diameter (i.e. 36 ± 15 nm maximum Feret diameter), potentially attributable to recently discovered crystal lattice strain effects recently reported by Quintanilla et al (2022) [46], but this is not verified. Overall, the PVPassisted route is an attractive method for synthesis of UCNPs with various luminescence emission properties, albeit a method which has not been well utilized by the wider research community.
To-date, the influence of key variables in the PVPassisted route of UCNP synthesis have not been established. The majority of studies have used a single PVP molecular weight (i.e. PVP40, with a molecular weight of 40,000 g mol −1 ) and (a seemingly arbitrary) fixed quantity of PVP (∼574 mg/14 μmols). The singular exception known to us is the study by Meng et al (2012), who report using 500 mg of PVP K-30 (molecular weight of 58,000 g mol −1 ), i.e. 8.6 μmols of PVP K-30,to produce NaYF 4 nanoparticles with unspecified diameter distributions [29]. We hypothesised that altering the amount and molecular weight of PVP in UCNP synthesis would influence UCNP morphology [45]. As such, the aim of this initial exploratory study was to establish what influence of PVP molecular weight and PVP quantity may have on UCNPs produced via this PVP-assisted low-temperature open-air synthesis method. For this initial exploratory study, we reduced the amount of PVP in the synthesis to an arbitrary value of 10%, i.e. an order of magnitude less than the nominal amount. As a point of comparison, Xia et al (2011) [2] found that a similar reduction of PVP amount to 10% nominal values was a critical threshold for producing Ag nanoparticles with different morphologies [2]. We selected NaYF 4 :Yb,Er UCNPs as our model nanoparticle system because it is a commonly used and well-studied UCNP composition, which minimises the variables involved and assists in inter-study comparisons.

Synthesis of UCNPs
PVP-UCNPs were produced as per a recent study by our group [17], which in-turn was adapted from prior studies [3,5,6] Synthesis required the preparation of two mixtures. For mixture A, appropriate amounts of each Ln 3+ stock solution (typically ∼2 ml each) were then combined in a 50 ml beaker to create a solution containing a total of 1 mmol of Ln 3+ ions of various ratios (see table 1). This solution was stirred vigorously for one minute and then heated to 120°C using a hotplate/stirrer to evaporate the aqueous content. This was deemed complete when the transparent residue (consisting of lanthanide nitrates [3,17]) at the bottom of the beaker stopped bubbling. 8 ml of ethylene glycol was then added, and the solution heated to 80°C whilst being stirred for 30 min Whilst maintaining temperature at 80°C, 58.5 mg (1 mmol) of NaCl was then added directly to the solution and stirred for a further 5 min 2 ml of the appropriate PVP stock (see  full power. In all instances the temperature of the solutions was measured directly by placing the hotplate/ stirrer thermocouple directly in the reaction mixture. The solution was maintained at 160°C for 2 h, during which UCNPs formed, with the reaction progressing from opaque white to an opaque yellow/orange colouration, likely due to formation of NO 2 from decomposition of nitrates. The UCNPs were then pelleted by centrifugation at 10,000 relative centrifugal force (RCF) for 30 min; the supernatant was removed and replaced with 10 ml of fresh EtOH; the UCNPs were then resuspended by sonication for 15 min. Subsequently two more wash steps were conducted as per the first wash step; UCNPs were then suspended in a final volume of 5 ml EtOH and stored at room temperature in sealed 50 ml falcon tubes.

UCNP characterisation 2.2.1. Quantifying sample weight
The dry weight of UCNPs produced in each sample was estimated by first pre-weighing empty open glass vials with a high precision (0.1 mg) balance five times (to quantify random error), then sonicating the UCNP samples for 10 min prior to transferring 500 μl of each UCNP sample into the vials within a ventilated fumehood, and subsequently allowing the EtOH to evaporate naturally. The vials were then weighed five times. Uncertainties were extrapolated from the standard deviation of five measurements combined with manufacturer's quoted random error of the pipette used (Gilson Pipeteman ® Classic P1000), i.e. ±1 μl.

Electron microscopy and UCNP diameter analysis
For scanning transmission electron microscopy (STEM), the as-prepared samples were sonicated for 10 min, then 20 μl of each UCNP sample was added to 1.98 ml of isopropyl alcohol and sonicated for 1 min. For each sample, a fresh grid (holey carbon film on 200 mesh, Agar Scientific Ltd) was gripped using tweezers and then dipped into the UCNP solution and the grid was allowed to dry via evaporation at room temperature. Samples were stored in a grid box. Each individual sample was then placed onto a single sample stage and affixed using conductive silver paint. Each sample was imaged in STEM mode using a variable pressure field emission gun scanning electron microscope (FEI Quanta 250) operating in high vacuum mode with an accelerating voltage of 30 kV at a working distance of ∼6 mm, at 100,000X magnification, and using a two-segment solid-state detector holder operating in bright-field mode, mounted under the sample. The diameter of UCNPs was estimated by importing STEM images acquired at 100,000X magnification into ImageJ [47,48] and manually measuring the maximum and minimum Feret diameters of ∼80-120 randomly selected UCNPs in each sample.
For scanning electron microscopy (SEM), the asprepared samples were sonicated for 10 min then 20 μl of the as-prepared UCNP samples was added to 1.98 ml of isopropyl alcohol and sonicated for 1 min. Silicon wafer fragments were placed on a 50°C hotplate and washed with 20 μl of isopropyl alcohol which was allowed to evaporate. Once dry, a single drop (∼10 μl) of the diluted UCNP sample was placed on the silicon wafer fragment and allowed to evaporate. Once dry, the silicon wafers were mounted to aluminium specimen stubs using adhesive carbon tabs and stored in a stub box. The stubs were then mounted on a multi-stub sample holder and imagined using the aforementioned SEM operating in high vacuum mode with an accelerating voltage of 10 kV, at a working distance of ∼6 mm, at 100,000X magnification, and using the standard secondary electron (SE) detector. For both STEM and SEM, images were saved in uncompressed .TIF format at a resolution of 1024 × 943 pixels.
Further high-resolution transmission electron microscopy (HR-TEM) of selected samples was performed using a Tecnai TF 20 X-TWIN microscope (FEI). Elemental mapping analysis was performed using an energy dispersive spectrometer (EDS r-TEM, EDAX) under an accelerating voltage of 200 kV. Samples were prepared as per STEM imaging.

Structure analysis via transmission Kikuchi diffraction (TKD)
Transmission Kikuchi diffraction (TKD), also referred to transmission electron backscatter diffraction (t-EBSD) [49,50] was performed in a field emission gun SEM (FEI Versa 3D) equipped with an Oxford Instruments Symmetry S2 EBSD detector. For TKD, the samples were prepared on a carbon mesh grid using the same steps as described above for the STEM measurements. The mesh grid was mounted at −60°o n custom-built TKD holder, as described in Tokarski et al (2021) [51]. The stage was tilted by 40°to result in sample tilt of −20°which is typical for off-axis TKD measurements. TKD measurements were carried out at an acceleration voltage of 30 kV using EBSD camera settings of 'Speed 1' (622 × 512 pixels) or 'Speed 2' (156 × 128 pixels) and a gain setting of 2. TKD maps were acquired with a step size of 10 nm. Oxford Instrument's Refined Accuracy method was used for indexing the TKD patterns with increased angular resolution. Subsequent, TKD analysis was carried out using the Matlab-based toolbox MTEX [52], which was also used to produce TKD maps. Dark field STEM images were recorded in forescatter electron detector (FSD) geometry using solid state diodes placed around the EBSD detector screen.

Structure analysis via powder x-ray diffraction (PXRD)
Dry powder samples were prepared by decanting ∼10 mg of UCNPs in EtOH into an open clean glass vial inside a ventilated fume hood and allowing the EtOH to evaporate naturally at room temperature.
Afterwards, the dry residue was scraped with a clean spatula within the vial to form a white powder. The samples were then placed into a Kapton-backed multiwell sample holder. PXRD measurements were obtained using a D8 Discover (Bruker Corporation) x-ray diffractometer using a Cu Kα x-ray source (λ = 1.5406 Å) and scanning 2ϴ angles between 10°a nd 90°in 0.017°increments at a rate of 0.5 s per increment. After acquisition, the background of each PXRD pattern was removed using the background subtraction function of Diffrac.eva software (Bruker Corporation). PXRD peaks were indexed by comparison to reference pattern data. The position of each peak was calculated using custom-written Matlab script (Matlab 2021b, Mathworks). The cubic (αphase) crystal lattice parameter, a, was calculated from each peak by: where l is the wavelength of the x-ray source, d hkl is the interplanar spacing associate with a given PXRD peak, hkl are the Miller indices corresponding to each peak, and 2ϴ is the two-theta angle of each peak. Uncertainties in a were quantified by the standard deviation of eight estimates of a from eight PXRD peaks.

Results and discussion
STEM is used for visualising the core crystalline structure of the UCNPs (see figure 3) whereas SEM imaging visualises the PVP coating around the UCNPs (see figure 4). The UCNPs produced were typically pseudo-spherical or pseudo-cuboid with varying diameters; these results are largely in-line with prior studies using the PVP-UCNP synthesis method [17].
Notably, a small number of unexpected rhombusshaped UCNPs were observed in the 10% PVP10 Er 3+ and 10% PVP40 Er 3+ samples (see figure 5). HR-TEM images (shown in figure 6) indicated that the rhombus UCNPs appear to be flat, indicated by lower contrast than rhombus UCNPs compared to pseudo-spherical UCNPs. Further, that a crystal lattice can be observed in both pseudo-spherical and rhombus-shaped UCNPs. SEM images acquired at a 45°sample tilt may support the hypothesis that these rhombus-UCNPs are flat and not bipyramidal (see Supplementary  figure 1), but these imaging experiments do not conclusively reveal the 3D morphology of the rhombus-shaped UCNPs. To further, investigate the crystal structure of individual nanoparticles, TKD measurements were performed on individual and clusters UCNPs. The inverse pole figure (IPF) map along the Z direction (which is the normal direction of the copper mesh) of a cluster of UNCPs (see figure 7(b)) indicates that all UCNPs were randomly orientated. Closer inspection of the diffraction patterns of rhombusshaped (figures 7(c) and (d)) and pseudo-spherical UCNPs (figures 7(e) and (f)) shows that they were all of cubic (α-phase) crystalline structure. EDS analysis (see figure 8 and Supplementary figure 3) verified that all  nanoparticles produced contain Na, Y, F, Yb, and Er, as expected for the desired UCNPs. Therefore, the elemental composition and structure of individual rhombus-shaped UCNPs produced herein is consistent with UCNPs of other morphologies. The synthesis of rhombus-shaped NaYF 4 :Yb,Er UCNPs is notable for two reasons. Firstly, to the best of our knowledge, UCNPs with rhombus-like shapes have exclusively been observed in studies of LiYF 4 host-lattice UCNPs (where they may be referred to as 'bipyramidal' or 'octahedral' morphology, whether or not definitive 3D imaging has occurred). In such studies, populations of UCNPs are often exclusively rhombus/bipyramidal morphology, not a mixture of shapes. Hong et al (2017) attribute this rhombus morphology to UCNP formation via the tetragonal unit cell {101} plane [53]. Notably, these other rhombus UCNPs are grown via methods requiring high pressures and/or inert atmospheres [30,[53][54][55][56][57][58][59][60][61][62][63]. In contrast, the PVP-UCNP synthesis method requires much simpler and lower-cost equipment. However, only a small fraction of the UCNPs observed in 10% PVP10 and 10% PVP40 samples were rhombusshaped, indicating multiple competing UCNP growth processes in these preparations. With the information available in this exploratory study, we are not able to elucidate a growth mechanism for the rhombusshaped UCNP morphology with the information currently available. It is likely that parameters such as reaction time, reaction temperature, reaction pressure, and solvent choice will influence the production of rhombus-shaped PVP-nanoparticles in addition to the aforementioned affects of PVP molecular weight and the quantity of PVP used. It is not clear why rhombus UCNPs form for 10% PVP10 and 10% PVP40 samples, but not the 10% PVP55 sample, however Xia et al (2011) note that longer chain PVP (i.e. PVP55) may less efficiently adsorb and pack onto nanoparticle  surfaces due to steric effects [2]. Control of UCNP morphology is particularly important for life sciences and biosensor applications, because (a) UCNP morphology determines uptake and translocation rates [64,65], (b) the shape of UCNPs influences their Förster resonance energy transfer (FRET) response properties [66][67][68], and (c) the sharp edges of rhombus morphology may result in localised photothermal heating effects [55]. Further research is required to determine if rhombus-only UCNP populations can be produced by the PVP-assisted method.
Diameter measurements (see figure 9), revealed that incorporation of different PVP molecular weights in synthesis at nominal quantities resulted in only minor differences in UCNP diameter, all producing UCNPs with diameters in the range of ∼55 nm. In contrast, reducing the quantity of PVP in the synthesis to 10% the nominal amount had a significant effect: the 10% PVP10 Er 3+ and 10% PVP40 Er 3+ UCNP samples, which were larger UCNPs on average with a dramatically wider spread of diameters than the samples using the nominal amount of PVP; i.e. 75 ± 24 nm and 93 ± 32 nm respectively, compared to an average of ∼ 53 ± 11 nm for comparable samples. The exception to this is the PVP 55 Er 3+ samples, which produced similar size distributions for both nominal and 10% PVP quantities. These results indicate that the quantity of PVP incorporated into synthesis must have a crucial role in controlling the morphology and diameter of UCNPs produced via the PVP-assisted route, e.g. an abundance of PVP may be necessary to control UCNP growth via a mechanism such as a templating effect, sequestering of precursors, or alteration of nucleation conditions. However, it is not clear from this exploratory study what mechanism(s) are involved and so carefully designed further studies are required to fully elucidate the role(s) PVP plays UCNP synthesis.
Upconversion luminescence spectra exhibited by Er 3+ UCNP samples (see figure 11 and Supplementary figure 2) were generally in accordance to that reported by Li and Zhang (2006) [3], where red and green emission ratios close to parity was obtained. The 10% PVP55 Er 3+ UCNPs exhibited the somewhat brighter emission relative to the other samples, although there is no clear reason for this slight enhancement in luminescence. Factors such as morphology-dependent solvent quenching, unit cell volume, and UCNP order could feasibly play a role, but no clear trend is evident [45]. The PVP40 Er 3+ UCNP samples provides a notable opportunity for direct inter-study comparison.
Notably the value of crystal lattice parameter, a, obtained for PVP40 Er 3+ was identical to those produced by Sikora et al 2013 [6] and MacKenzie et al 2022 [17] (i.e. 5.52 Å), indicating excellent inter-study repeatability. However, the PVP40 Er 3+ UCNPs produced in this study exhibited a relatively low red/green emission ratio of 0.77, which is considerably less than the red/green ratio achieved in MacKenzie et al (2022) and Sikora et al (2013) (i.e. approximately ∼3) [6,17]. Furthermore, the PVP40 Er 3+ UCNPs produced in this study were significantly larger than those produced by MacKenzie et al (2022) (i.e. 52 ± 10 nm versus 42 ± 11 nm respectively) [17]. Whilst there are trivial reasons for inter-study differences, e.g. operator variability, differences in glassware or heating medium, the most likely reason for these inter-study difference to have arisen is use of a different hotplate stirrer apparatus, which will have imbued a different heating rate to the prior study. MacKenzie et al used an IKA RCT basic hotplate/stirrer and this study used a Fish-erbrand™ AREX with VTF hotplate/stirrer. Notably, both these hotplates have different temperature-cotrol logic which cannot be matched. Heating rate is known to be an important parameter in UCNP synthesis [69,70]. Indeed, a recent exemplary study on reproducible UCNP synthesis by Andresen et al identified heating rate as a key parameter [34]. We hypothesise that heating rate may affect order within UCNPs, therefore influencing UCNP optical properties even where otherwise identical unit-cell parameters have been obtained.
From this exploratory study, it is not clear if the rhombus-shaped nanoparticles have different properties from the pseudo-spherical nanoparticles produced. Discerning the optical behaviour of individual UCNPs require upconversion studies of single nanoparticles, which would necessitate super-resolution optical microscopy approaches, however it is not clear if such approaches could discriminate between rhombus and pseudo-spherical UCNPs [71]. Therefore producing populations of exclusively rhombus-shaped UCNPs would be most desirable for straight-forward understanding of the optical properties of rhombusshaped PVP-UCNPs. Notably, overall upconversion emission intensity from PVP40 Tm 3+ UCNPs was much stronger than any of the Er 3+ UCNPs, with strong blue (∼474 nm) and strong red emission (∼660 nm) (see figure 11). Whilst the PVP40 Tm 3+ UCNPs were around 11% larger in diameter than PVP40 Er 3+ UCNPs, they displayed a somewhat reduced crystal lattice parameter, a, of 5.502 ± 0.03 Å compared to 5.524 ± 0.001 Å for the PVP40 Er 3+ UCNPs. Furthermore, ICP-MS revealed that the PVP40 Tm 3+ UCNPs incorporated a slightly reduced amount of photonically inert Y 3+ ( 3.6 relative % less) and a slightly increased amount of photonically active Yb 3+ (3.2 relative % more) compared to the average of Er 3+ containing UCNPs (see table 3). It is not clear why the PVP40 Tm 3+ UCNPs incorporated proportionally more Yb 3+ , but other studies have noted that increasing Yb 3+ can increase emission intensity [6]. Therefore increased upconversion emission from the PVP40 Tm 3+ UCNPs could possibly be explained by a greater density of Yb 3+ ions and reduced inter-ion separation [6,72]. When plotted on a CIE 1931 chromaticity diagram, the PVP Tm 3+ UCNPs yield apparent purple emission. This contrasts results by Li and Zhang (2006), whose PVP40 Tm 3+ UCNPs exhibited strong blue and relatively weaker red emission (∼650 nm) [ 1 G 4 → 3 F 4 and 3 F 3 → 3 H 6 emission transitions]. It should also be noted that the PVP40 Tm 3+ UCNPs may also exhibit luminescence ∼800 nm which could not be recorded with the instrumentation used in this study [23]. Overall, these results demonstrate that PVP40 Tm 3+ UCNPs with strong dual-band blue and red emission properties can be produced via the PVP-assisted route.
The limitations of this study should be acknowledged. Firstly, only three PVP molecular weights were tested. A wider range of PVP molecular weights are commercially available, but these have different bulk properties (e.g. higher molecular weight PVP samples are poorly soluble in ethylene glycol), and so would have required modification of the experimental protocol (e.g. heating to aid polymer solubility) which is not conducive for this initial comparative study. Additionally, hybrid mixtures of different PVP molecular weights and hybrid solvent compositions were not explored. Secondly, only two quantities of PVP were investigated: the nominal amount (574 mg) and 10% of the nominal amount (57.4 mg); different quantities of PVP may yield differing results and should be explored in future. Additionally, the longterm stability of PVP-UCNPs has not been studied [73,74]. Thermogravimetric analysis may be useful for studying the relative amount of PVP contained in each UCNP sample [1]. Thirdly, this work was conducted using lanthanide oxide starting materials; prior work indicated that using other starting materials (particularly lanthanide nitrates) may disrupt nanoparticle formation, perhaps due to varying nucleation interactions with the PVP [17]. Solution-state studies of directly excited lanthanide luminescence may be required to study the relevant processes involved.

Conclusions
In this exploratory study, we have demonstrated that altering PVP molecular weight and PVP amount used in synthesis can influence the morphology and diameter of cubic (α-phase) NaYF 4 :Yb,Er UCNPs produced via a modest-temperature open-to-air synthesis route. The choice of PVP molecular weight (i.e. 10,000, 40,000, 50,000 kDa) had minimal effect on resulting UCNP size distributions when incorporated in synthesis at nominal amounts. However, reducing the quantity of PVP10 and PVP40 to 10% of the nominal amount produced much wider size distributions of UCNPs, with a notable sub-population of rhombus-shaped UCNPs. The exception to this were the samples incorporate ng PVP55, where the difference arising from varying PVP amount was marginal.
Notably, a small fraction of rhombus-shaped NaYF 4 :Yb,Er UCNPs were produced in the 10% PVP40 Er 3+ and 10% PVP10 Er 3+ samples (∼11% and ∼4% of total UCNPs respectively). Single nanoparticle structural and elemental analysis supports the conclusion that these rhombus-shaped nanoparticles were cubic (α-phase) NaYF 4 :Yb,Er UCNPs. Rhombusshaped UCNPs were not observed in PVP55 samples. Despite these intriguing results, the mechanism for how PVP influences the growth of UCNPs has not been elucidated by this exploratory study, with templating and precursor sequestering being plausible effects. Ultimately, further work is required to fully understand and exploit the potential of PVP in UCNP synthesis.
In broader context, the PVP-assisted route of UCNP synthesis is promising but remains largely unexploited despite it's inherent advantages of modest reaction temperatures, open-to-air synthesis, and simple equipment requirements. The full range of UCNP morphologies and photonic emission properties that can be produced by altering PVP parameters remains to be established, with critical reaction parameters of synthesis temperature, synthesis pressure, reaction time, reaction atmosphere, and solvent choice yet to be explored. To-date, the PVP-assisted UCNP synthesis route has produced UCNPs with blue and red emission (NaYF 4 :Yb,Tm), green/red emission (NaYF 4 :Yb,Er), and strong red emission (NaYF 4 :Yb,Er,Mn). Further investigation of host lattice, lanthanide, and transition metal dopant combinations could enhance emission properties and unlock further desirable emission wavebands. Indeed, the potential for scaled-up production and multi-colour UCNP populations indicates that the PVP-assisted route could be particularly useful for large-scale UCNP applications, such as display technologies and stochastically-patterned nanosecurity inks.  Finally, we would like to thank the two anonymous peer-reviewers whose thorough contribution improved this manuscript considerably. The authors have declared that no conflicting interests exist.

Data availability statement
The data that support the findings of this study are openly available at the following URL/DOI: https:// doi.org/10.6084/m9.figshare.22434913.v1.

Author contributions
• R B Synthesised all UCNP samples, contributed to experimental design and methodology, conducted preliminary analysis of PXRD, SEM, and upconversion luminescence data.
• J B Conducted the TKD measurements and associated analysis, created Figures, and contributed to writing of the manuscript, and reviewed the manuscript.
• M G Conducted high-resolution TEM imaging and EDS measurements.
• G C Conducted the TKD measurements and reviewed the manuscript.
• R P Conducted upconversion luminescence emission measurements and reviewed the manuscript.
• L M Secured funding, conceived the project and methodology, conducted STEM and SEM imaging, analysed the data, produced figures, curated data, and wrote the manuscript.
• All authors gave final approval for publication.